t-test de triplicats dans une trame de données horizontale

Question

J'essaie de faire une boucle, donc je peux tester plusieurs conditions: cond_A, cond_B et cond_C, chacune contre le même contrôle ('ctrl'). Chaque condition et contrôle est représenté par un triplicat. Comme résultat je voudrais obtenir un dataframe avec des noms de condition et pvalues.

Voici mon entrée:

structure(list(ctrl_1 = 1L, ctrl_2 = 2L, ctrl_3 = 3L, cond_A_1 = 4L, 
    cond_A_2 = 4L, cond_A_3 = 4L, cond_B_1 = 5L, cond_B_2 = 5L, 
    cond_B_3 = 7L, cond_C_1 = 8L, cond_C_2 = 9L, cond_C_3 = 2L), .Names = c("ctrl_1", 
"ctrl_2", "ctrl_3", "cond_A_1", "cond_A_2", "cond_A_3", "cond_B_1", 
"cond_B_2", "cond_B_3", "cond_C_1", "cond_C_2", "cond_C_3"), class = "data.frame", row.names = c(NA, 
-1L)) 

et les résultats escomptés avec pValues ​​hypothétiques:

cond_A_pval cond_B_pval cond_C_pval 
0.05 0.9 0.006 

Voici mon point de départ:

pval<-apply(df,1,function(x) {t.test(x[1:3],x[4:6])$p.value}) 

Answer 1

Effectuez les opérations suivantes:

df <- structure(list(ctrl_1 = 1L, ctrl_2 = 2L, ctrl_3 = 3L, cond_A_1 = 4L, 
       cond_A_2 = 4L, cond_A_3 = 4L, cond_B_1 = 5L, cond_B_2 = 5L, 
       cond_B_3 = 7L, cond_C_1 = 8L, cond_C_2 = 9L, cond_C_3 = 2L), 
       .Names = c("ctrl_1", "ctrl_2", "ctrl_3", 
          "cond_A_1", "cond_A_2", "cond_A_3", 
          "cond_B_1", "cond_B_2", "cond_B_3", 
          "cond_C_1", "cond_C_2", "cond_C_3"), 
       class = "data.frame", row.names = c(NA, -1L)) 

library(tidyr) 

# Reshape the data into key-value pairs. 
# It is generally advisable to have data in tidy format. 
df <- gather(df) 
# Remove the _1, _2, etc. 
df$group <- gsub("_\\d", "", df$key) 

#Now you can loop through the groups. Note that "ctrl" is the first group: 
sapply(unique(df$group)[-1], function(x){ 
    t.test(df[df$group == "ctrl", "value"], df[df$group == x, "value"])$p.value 
}) 

cond_A  cond_B  cond_C 
0.07417990 0.01477836 0.17957429 

Voir aussi Looping through t.tests for data frame subsets in r